The major argument for promoting the development of spin electronics is the low power dissipation. The aim of the thesis is to determine and optimize the power efficiency of these devices. We focus the study on the power dissipated by two kinds of devices. On the one hand, the devices allowing the reversal of the magnetization of a magnetic layer by a transverse spin current, namely the Spin-Orbit Torque effect (SOT), and on the other hand the devices based on topological materials.

In this context, the definition of useful power - or efficiency - is an open problem. Indeed, the thermodynamics of this type of non-equilibrium system involves cross-effects between the degrees of freedom of the electric charge carriers, those of the spin of these carriers, as well as those of the magnetization of the adjacent layer.

We have developed a variational method in order to establish the stationary state of a Hall bar and the power dissipated in a load circuit. Preliminary measurements have recently validated the prediction in the case of the anomalous Hall effect. The project aims to generalize the study to SOT and topological materials.